Intraoperative acquisition of DTI in cranial neurosurgery: readout-segmented DTI versus standard single-shot DTI

3D color multimodality fusion imaging as an augmented reality educational and surgical planning tool for extracerebral tumors

Extracerebral tumors often occur on the surface of the brain or at the skull base. It is important to identify the peritumoral sulci, gyri, and nerve fibers. Preoperative visualization of three-dimensional (3D) multimodal fusion imaging (MFI) is crucial for surgery. However, the traditional 3D-MFI brain models are homochromatic and do not allow easy identification of anatomical functional areas. In this study, 33 patients with extracerebral tumors without peritumoral edema were retrospectively recruited. They underwent 3D T1-weighted MRI, diffusion tensor imaging (DTI), and CT angiography (CTA) sequence scans. 3DSlicer, Freesurfer, and BrainSuite were used to explore 3D-color-MFI and preoperative planning. To determine the effectiveness of 3D-color-MFI as an augmented reality (AR) teaching tool for neurosurgeons and as a patient education and communication tool, questionnaires were administered to 15 neurosurgery residents and all patients, respectively. For neurosurgical residents, 3D-color-MFI provided a better understanding of surgical anatomy and more efficient techniques for removing extracerebral tumors than traditional 3D-MFI (P < 0.001). For patients, the use of 3D-color MFI can significantly improve their understanding of the surgical approach and risks (P < 0.005). 3D-color-MFI is a promising AR tool for extracerebral tumors and is more useful for learning surgical anatomy, developing surgical strategies, and improving communication with patients.

Classification Algorithms for Brain Magnetic Resonance Imaging Images of Patients with End-Stage Renal Disease and Depression

This study was aimed to explore the relationship between depression and brain function in patients with end-stage renal disease (ESRD) complicated with depression based on brain magnetic resonance imaging (MRI) image classification algorithms. 30 people in the healthy control group and 70 people in the observation group were selected as the research objects. First, the preprocessing algorithms were applied on MRI images. With the depression classification algorithm based on deep learning, the features were extracted from the capsule network to construct a classification network, and the network structure was compared to obtain the difference in the distribution of brain lesions. Different classifiers and degree centrality, functional connection, low-frequency amplitude ratio, and low-frequency amplitude were selected to analyze the effectiveness of features. In the deep learning method, the neural network model was constructed, and feature extraction and classification network were carried out. The classification layer was based on the capsule network. The results showed that the correct rate of the deep learning feature extraction network structure combined with the capsule network classification was 82.47%, the recall rate was 83.69%, and the accuracy was 88.79%, showing that the capsule network can improve the heterogeneity of depression. The combination of fractional amplitude of low-frequency fluctuation (fALFF), DC, and amplitude of low-frequency fluctuation (ALFF) can achieve the accuracy of 100%. In summary, MRI images showed that patients with depression may have neurological abnormalities in the white matter area. In this study, the classification algorithm based on brain MRI images can effectively improve the classification performance.